In order to avoid the temperature cross-sensitivity during refractive index measurement, a sensor based on a superimposed long period gratings fabricated in few mode fiber was proposed. The superimposed long period grating was inscribed under the CO2 laser irradiation as following: first, a long period fiber grating with a period of 654 μm and length of 30 mm was inscribed, then the few mode fiber was rotated 180° to inscribe second long period fiber grating with a period of 819 μm and the same length. Two resonance dips at 1 487.2 nm and 1 533.0 nm appeared in the transmission spectrum, which were formed by coupling between different guided modes. The difference of two resonance dips increases with the refractive index while is constant with temperature increase, which can eliminate temperature cross-sensitivity during refractive index measurement. The experimental results show the temperature sensitivity of the sensor is -0.002 5 nm/℃ in 20~70 ℃ temperature range and the refractive index sensitivity of 143 nm/RIU in 1.333 3~1.376 6 refractive index range. The temperature sensitivity of the sensor is much lower than the refractive index sensitivity, which means that it is insensitive to temperature. Compared with traditional optical fiber sensors, the sensor has the advantages of small temperature cross-sensitivity, high refractive index sensitivity, small size and compact structure, and it can be widely used in industry, water conservancy, medicine and other fields.

The performance degradation of fiber Bragg grating strain sensors is unavoidable under stress fatigue, in order to assess the ultimate sensing life of fiber Bragg grating sensors, the full scale relative error expression of stress fatigue effect is deduced from the principle of fiber Bragg grating, and the industrial instrument accuracy class is used as the limit state threshold of the sensor. And then a method based on the beam of constant strength is put forward to test the ultimate sensing life. The ultimate sensing life of fiber Bragg grating in high pile wharf structural health monitoring application is assessed by this method, the failure threshold of full scale relative error is set as 4%, 4 fiber Bragg grating sensors suffer from 132 million times alternate strain, when the fatigue number reach to 132 million times, 2 sensors reach the ultimate sensing life, 1 sensor proximity reach the limit. That is, the maximum sensing life is 132 million times. Experimental results show that the proposed method can effectively test the ultimate sensing life of stress fatigued fiber Bragg grating strain sensors in different structural health monitoring application.

The fabrication mechanism of Long-Period Fiber Gratings(LPFGs) in standard SMF-28 telecommunication fibers without hydrogen loading were explored by using direct writing method based on femtosecond laser pulses, with pulse duration of 100 fs, 1kHz repetition rate and a central wavelength of 800 nm. The experimental system to write LPFGs is adopted directly the horizontal and vertical double CCD video to realize monitoring and alignment of fiber. The LPFGs with different spectral characteristics were fabricated by adjusting the number of grating period, energy density of laser irradiation and duty ratio. The result shows that resonance peak loss generate single-peak when the period of grating is constant, and the variations of the number of grating periods and pulse energy of laser irradiation will lead to the change of resonance peak loss. The variations of duty ratio will lead to the generation of resonance rejection band of LPFGs from single-peak to multi-peak plus lesser out-of-band loss. And finally, a resonance transmission spectra with 15 dB attenuation, 3 dB bandwidth of 15 nm, and less than 2 dB out-of-band losses at wavelength of 1 540 nm was obtained.

Three dimensional deformation field of the dumbbell shaped specimen surface is obtained by color digital holographic system. The young′s modulus of TC4 titanium alloy is calculated by the three dimensional deformation field in the linear zone of the specimen, which is 90.79 GPa. Moreover, the young′s modulus of the material is measured as 99.05 GPa by the strain measuring instrument. The experimental results is approximate to the results of traditional strain measuring instrument. It indicates that the method of measuring Young′s modulus by color digital holography is feasible and effective. The method has the characteristics of real-time recording, full-field, non-contact and so on. It can adapt to the real-time detection in the process of material service under high temperature environment.

In order to correct phase aberrations in the digital holographic microscopy measurement system, a numerical method combined with physical method, namely reference lens method is applied in off-axis digital holographic microscopy system. Firstly, the frequency spectrum of digital hologram is analyzed. By filtering and displacing a part of spectrum, the first-order phase aberration is corrected. Then, the reference lens is introduced in reference path. The axial best position of reference lens is discriminated according to the spectrum center energy. By using this physical method, high-order phase aberration is corrected. Based on the analysis of the phase correction theory, the validity and accuracy of the reference lens method is verified by the combined experiment. The experimental results show that the method of reference lens can accurately correct phase aberration. The standard deviation can reach 0.8 nm. And the measured results agree with measurement results of mechanical profiler, showing that reference lens method can effectively and accurately correct phase aberration of off-axis digital holographic microscopy system.

A new method for geometric features extraction of ship target in high-resolution Synthetic Aperture Radar (SAR) image was proposed, After detecting and locating ship targets from high-resolution SAR images. The algorithm continues to acquire target slices to construct the process of the ship target geometric feature extraction. Firstly, the algorithm obtained a saliency map, completed the detection and positioning of ship targets, and obtained the ship target slices. Secondly, the algorithm extracted the geometric features based on the resulting ship slices. The slices was estimated by the azimuth to obtain the exact minimum bounding rectangle, then effective and accurate extraction of geometric features can be completed. Finally, the algorithm appied to SAR image target detection, which is efficient proved by experimental results. The experiments on TerraSAR-X and a large number of satellite data demonstrate that the proposed algorithm can extract the geometric features with high accuracy and good stability. Unlike traditional methods, the use of improved spectral residual visual significant computational models to locate and segment ship targets can effectively reduce the false alarm rates, and the detection speed increased by 25% to 50%. And it is suitable for practical requirements of ship target detection in high-resolution images.

To solve the inaccurate problem of global motion estimation due to local motion in video stabilization, a real-time stabilization method was proposed. The method proposes an approach of feature points iterative filtering algorithm based on minimal spanning tree and applies spanning tree similarity of successive frames to measure feature matching and discard wrong matched points and points in moving foreground. Then, an adaptive weighted method was applied to correct the transformation matrix to solve the wobble problem due to few feature points caused by foreground occlusion. Finally, an modified dual Kalman filters based on motion queue was proposed to correct measurement noise covariance adaptively and adjust smoothess of filter dynamically. It can process videos with intended camera motion and random jitters effectively. Experimental results show that the proposed method can perform well even in situation of local foreground motion and intended camera motion. Under circumstance of Intel Core i5 3.30 GHz CPU, the proposed method can stabilize videos of 640×360 resolution at a rate of 40FPS. And it has real-time advantage since next frame information is not needed during computation.

In order to improve the calibration reliability of laser 3D projection calibration system and to obtain the best calibration parameters, according to the basic principle of the system, a method is proposed to combine the laser 3D projection observation model with the calibration parameter solution model to build a laser 3D projection calibration model approach. The method uses two steps to solve the calibration parameter of laser 3D projection calibration system. The first step is to establish the solution model of laser 3D projection observation by the isometric method, and to introduce the solution of the initial value of the model. The design results of this initial value avoid the drawbacks of the Newton iterative method. In the second step, the quaternion method is used to establish the model of the calibration parameters. The simulation results show that the error of the 3D projection calibration system is less than 0.3 mm when the mean value of the normal distribution is 0 and the standard deviation is 0.5″.At the same time, the laser tracker system is used as a benchmark to verify the calibration precision and reliability of the method. The results show that the maximum error is within 0.4 mm in the range of 3 to 5 m, which meets the projection of laser 3D projection system positioning accuracy. The calibration model can realize the calibration of the laser 3D projection calibration system, and the calibration precision is high and the calibration method converges fast and improves the system calibration rate which provides a new algorithm for the development of laser projection system.

In order to avoid the influence of the interference noise introduced by the front and rear surfaces while measuring the parallel plate, an absolute measurement method for the optical homogeneity was proposed based on the principle of phase shifting by lateral displacement of the point source. Three steps of the proposed method are as follows: the front and rear surfaces of the parallel plate interference measurement, the transmission wavefront through the parallel plate measurement and the empty cavity of the interferometer measurement. The wavefront is recovered by four phase-shifting interferograms captured at the same time and the homogeneity distribution of the parallel plate could be calculated according to the measurement results. An optical parallel plate with a thickness of 60mm was tested on a vibration non-isolated platform and the measurement result indicates that the peak-to-valley value is ΔnPV=3.32×10－6 and root-mean-square value is ΔnRMS=2.63×10－7. Compared with the results obtained by the wavelength tuning interferometer, the deviation of the peak-to-valley value is ΔPV=5×10－7 and the deviation of the root-mean-square value is ΔRMS=－7×10－9. The measurement results are in good agreement with those obtained by wavelength tuning interferometer. The measurement accuracy of homogeneity obtained by the proposed method is 1×10－6 under vibration conditions.

Giant Steerable Science Mirror (GSSM) is largest flat mirror as the tertiary mirror of Thirty Meter Telescope (TMT).To evaluate the performance of GSSM under gravity load, the mirror figure needs to be tested. Erorr analysis is the most imortant step for the reliability of metrology. For its large size and plate character, Sub aperture stitching will be used to reach the full aperture figure. GSSMP is the prototype of GSSM. Research on GSSMP sub aperture testing error will help understanding and learning how to build GSSM. The stitching error is divided into two parts: rigid body location error and middle frequency turbulence. According to each part, theoretical analysis and experience have been done drawing the conclusion that the algorithm with probable marks will suppress rigid body error efficiently and the influence of the turbulence is achieved by smaller aperture system and interferometer. What is more, stitching order is also related to stitching order. Using one reference aperture and fitting other apertures onto it allow introducing least error to the final result. All the performance is specified by Slope RMS,which is required by TMT.After fitting, Shift error is 10－6 μrad, Piston error is 10－6 μrad, Tip/tilt error is 10－6 μrad and Air turbulence is 0.04 μrad. The error involved by the rigid body is ignorable, and the air turbulence is 3nm in RMS. By Slope MS, the reqiured metric of TMT, the influence of stitching error will be discussed and the orders chosen here did not increase the Signal to Noise ratio significantly in the figure at the level of 2%. Here the conclution was drawed that feature will help a lot in lowering the stitching error.

High stability mode-locked fiber laser whose key is to obtain a stable mode-locked pulse output has broad applications in the optical frequency comb, micro-processing and other fields. In this paper, chirped fiber Bragg grating, as the spectral filter and the dispersion control element, has been used to control the spectral shape and chromatic dispersion of the Yb-doped fiber laser. The mode-locked polarization maintaining fiber laser was obtained by semiconductor saturated absorption mirror. The maximum output power of laser was 9.8 mW by the pump power of 52 mW. The fundamental repetition rate was 15.7 MHz. The center wavelength and the spectral bandwidth were 1 030 nm and 7.2 nm, respectively. The pulse width was measured to be 4.26 ps, corresponding ideal Fourier transform limit pulse width is about 150 fs.

A compact laser-induced breakdown spectroscopy using a low-energy, high-repetition rate Nd: YAG DPSS laser as excitation source has been developed. The influences of laser repetition rate and laser power on LIBS signal were analyzed. When the laser operated at 4 kHz, pulse energy of 745 μJ and pulse width of 17 ns, the seven emission lines of Lead (Pb) was obtained. Compared with the traditional single pulse LIBS, the number of emission lines is increase, also the signal-to-noise ratio is improved by almost 73%. Based on simplified setup and low energy, high-repetition rate laser-induced breakdown spectroscopy will be applied in more emerging fields.

A widely tunable narrow-linewidth photonic microwave generation scheme based on the period-one dynamics of an optically injected semiconductor laser combined with optoelectronic loops is proposed and experimentally demonstrated. The results show that the optical injection first drives the laser into period-one dynamics so that its intensity oscillates at a microwave frequency, which can be widely and continuously tuned in the range from 8 GHz to 67 GHz. A single sideband photonic microwave generation at 24.3 GHz is demonstrated with a proper injection condition. By using optoelectronic loops, the linewidth of the photonic microwave is effectively reduced from 8.6 MHz to 30 kHz and a signal to noise ratio above 40 dB is obtained.

A series of double-axis flexible support structure is put forward to ensure the surface figure and stability of the miniature space camera ultra-light reflectors. Taking the ultra-light mirror surface under multiple operating conditions as the goal, the integrated optimization method is used to optimize the support structure, and the statics of the optimized structure under the self-weight and temperature conditions are analyzed. The root-mean-square of each condition is within 3.5 nm, which is far better than the design requirement. The adhesive strength of the developed mirror assembly was checked and the dynamic performance was analyzed by finite element method and test. The bonding area of flexible support structure and mirror is 1 138 mm2. The basic frequency of the X, Y and Z directions of the mirror assembly is above 500 Hz. The relative error of the test and analysis results is less than 6.5%, which verifies the correctness of the finite element analysis model. At the same time, it shows that the series of double-axis flexible support structure design is reasonable, and the integrated optimization method is reliable.

An aspherical zoom optical system is designed to solve the problems of adjustment of internal and external coherence factors, lack of energy utilization and short life of module of NA 1.35 immersion lithography system. This zoom optical system consists of both three lenses groups and the stop as well as the image plane. The system consists of 7 pieces of lenses with pupil diameter of 42 mm, view angle of 1.89°, which can realize the focal length of 700 mm~1 830 mm, zoom ratio of 2.16, and the three aspheric surfaces are all designed in concave surfaces of optical elements. The design results show that the system is with good imaging quality of RMS diameter less than 40 μm and distortion less than 0.5%, which can satisfy the usability requirements of the immersion lithography lighting system.

A method is proposed for improving the stability of an Optoelectronic Oscillator (OEO) based on self-phase-locked technique. The output frequency stability of the OEO is improved by controlling an Optical Tunalble Delay Line (OTDL) according to the phase change of the OEO loop. The principle and noise transfer model are analyzed theoretically and verified experimentally. The experimental results show that the self-phase-locked OEO with a feedback control loop has a frequency drift within 0.9 ppm in 4 000 s, and its stability is greatly improved.

One-dimensional (1D) photonic crystal structure model was theoretically analyzed through analytical expressions with various structural parameters. The bandgap characteristics of the 1D photonic crystal resulting from the structural parameters were further simulated by a three-dimensional time domain finite difference numerical method. The results show that the bandwidth and center frequency of the 1D photonic crystal bandgap are significantly influenced by hole period and radius, InGaAsP waveguide width, refractive index of bottom cladding, and filling factor. A 1D photonic crystal nanobeam cavity is successfully designed for a Q-factor up to 104 with a designated resonant wavelength of 1 561.7 nm, showing the validation and practicability of the aforementioned results. The study of bandgap characteristics of 1D photonic crystal structure produces an important effect on the integrated photonic components such as formed lasers and modulators.

In order to explore the relationship between single crystal sapphire subsurface damage depth and polishing parameters of cluster magnetorheological finishing, a relationship model of the subsurface damage depth and the polishing parameters was established. The influence of the polishing parameters on the subsurface damage depth was studied, the relation model was verified by orthogonal experiments. White light interferometer was used as the measuring tool, α-Al2O3 was used as the polishing liquid, and three levels were selected for each experimental factor in the experiment. The experimental results show that the subsurface damage depth is related to abrasives size and polishing pressure, the subsurface damage depth increases with the increase of the polishing pressure and the abrasives size .The effect of the polishing pressure on the subsurface damage depth is much greater than that of the abrasives size. When the polishing pressure and the abrasives size are 25 kg and 280 nm, the minimum subsurface damage depth reach 0.9 nm after 100 minutes polishing, the polishing pressure is the main factor affecting the subsurface damage depth in cluster magnetorheological finishing. Under the current experimental conditions, the subsurface damage can be quickly removed when the polishing pressure is 25 kg.

Based on the theory of Surface plasmon polaritons, a plasmonic structure consisting of a metal-insulator-metal waveguide and a semi-ring stub was proposed to investigate the transmission properties by the finite-difference time-domain method. Simulation results show that there is a sharp and asymmetric Fano profile in transmission spectrum, which originates from the interference between a discrete state and a continuous state in the semi-ring stub. The tuning of the Fano profile is realized by changing the size of the semi-ring stub, and the sensitivity and figure of merit of this structure is 575 nm/RIU and 5 671. Moreover, multiple fano resonances are induced by setting a rectangular cavity in the waveguide, a figure of merit as high as 6 555, which offer flexibility in the design of the structure. The waveguide structure may find widespread applications in highly integrated optical circuits, special for nano bio-sensor.

A single photon avalanche diode with non-contact guard ring was proposed based on standard CMOS technology. The influences of the guard ring's spacing on the electric field distribution and the avalanche probability of the device were analyzed by using Silvaco Atlas. The dark count probability and the photon detection efficiency of SPAD operated under gate-model were calculated based on a physical model. The results shown that the device has optimal performance at the spacing d=0.6 μm. Under these conditions, a breakdown voltage of 13.5 V is obtained, and the dark current is as low as 10－11A. When the excess bias voltage is 2.5 V, the dark count probability is only 0.38%, the responsivities within the spectral range wavelength of 400 nm to 700 nm are decent, and the peak photon detection efficiency is up to 39% at 500 nm.

The wavelength/temperature coefficient of AlGaInAs MQWs FP laser devices are measured by analyzing the wavelength changed with ambient temperature at the same heating power. Also, the wavelength of device changed with heating power under RT are measured, and the thermal resistance of device is calculated to be 183K/W. Then, the temperature stress storage experiment are carried out on the device, the results showed: a slowly increased of wavelength happened as the ambient temperature increased from 120℃ to 220℃; and, a obviously blue shifted of wavelength happened while the temperature arrived at 225℃, the wavelength of device shifted to 1 265 nm from 1 297 nm before experiment; as the temperature reached to 235℃, the wavelength of device shifted to 1 258 nm, and the mode spacing decreased from 0.92 nm before experiment to 0.84 nm, that the effective refractive index of optical mode increased from 3.66 to 3.77; finally, the device became failure as the temperature arrived at 240℃. The possible reasons were mainly due to the Al, Ga and In atoms in the waveguide, quantum well and barrier layers of epi-wafer migration and the composition of these layers changing under the high temperature stress. The results will more provide the foundation for failure mechanism analysis and performances improvement of device under high temperature.

Compared to traditional RSOA, the reflective quantum dot semiconductor optical amplifier (R-QDSOA) has the picosecond carrier recovery rate and the modulation bandwidth of dozens of GHz. It is reveal that R-QDSOA has a great application prospect in high-speed WDM-PON. The simulation model of R-QDSOA is built according to the carrier rate equations and the light field transmission equations. And then the high-pass properties under the condition of gain saturation are investigated. The results show that the maximum signal gain and 3dB cutoff frequency can be improved by increasing the input optical power, maximum mode gain, active area length and reducing the injection current; the 3dB cutoff frequency of the R-QDSOA can achieve larger than 10GHz with reasonable parameter. This research makes R-QDSOA a promising candidate for re-modulation scheme of colorless WDM-PON and has theoretical significance for improving the modulation bandwidth of R-QDSOA.

The achromatic characteristic of polarization modulation module- double combo wedges modulator was studied, in order to improve the sampling accuracy of spectral polarization data and image in spatial modulating spectropolarimetry. According to the achromatic principle and the structure of the double combo wedges , the relationship between the first derivative of the phase retardance of the double combo wedges modulator and the birefringence and the wedge angle was deduced, which the double combo wedges modulator was composed by different birefringent crystal materials. The Matlab program was used to find the optimal combination. Based on the theory, the experimental contrast was carried on with KDP-Quartz and Quartz-Quartz combination .The experimental results show that the data of the first-order bright stripe of the KDP-Quartz combination is smaller than that of the Quartz-Quartz combination. The achromatic characteristic of KDP-Quartz combination is better than that of Quartz-Quartz combination, which is consistent with the theoretical analysis.

A method for generating hybridly polarized vector beams by manipulating Pancharatnam-Berry (PB) phase is proposed. According to the law of optical axis varying with the spatial coordinates, a PB phase element is used to manipulate the PB phase and generate vector beams with local linear polarization. When the vector beam is incident to a quarter-wave plate, a hybridly polarized vector beam is obtained. By measuring the Stokes parameter, the polarization distribution of the output light field is reconstructed. The experimental results show that the output hybridly polarized field contains all polarization states on the longitude of Poincaré sphere perpendicular to S3 axis when input wave is linearly polarized light with polarization angle θ0=0. The output hybridly polarized field contains all polarization states on the longitude of Poincaré sphere perpendicular to S3 axis when input wave is linearly polarized light with polarization angle θ0=0. When the polarization angle θ0=π/2, the output hybridly polarized field includes all polarization states on the longitude of Poincaré sphere perpendicular to S1 axis.

It is difficult to measure the particle size distribution for multi-modal particle system in dynamic light scattering technique, because of the ill posed problem in the data inversion. In this paper, the smooth constraint penalty term is added to the objective function of Tikhonov regularization, to improve the performance of inversion by enhancing the constraint to the solution. Three groups of simulated bimodal particle system (190/443 nm, 282/953 nm, 457/553 nm), one group of simulated unimodal particle system (564 nm), and one group of simulated trimodal particle system (292/591/889 nm), as well as two groups of experimental particle system (306/974 nm, 300/502 nm), were tested in this paper. The results show that, in the regularization inversion, adding the smooth constraint penalty term, can effectively eliminate the spike and spurious peaks in reversed particle size distribution, improve the resolution of particle size distribution and increase the ability of noise reduction, which, sequentially can make better use of the advantage that Multi-angle Dynamic Light Scattering (MDLS) technique could be able to provide more ultrafine particle size distribution information in medium and large size range, and then realize the accurate bimodal and multi-modal particle measurement from nanometer to submicron size range.

In order to study the propagation of near infrared light in the skin tissue, the human skin tissue model was established, including absorption coefficient, scattering coefficient, refractive index and anisotropy factor of epidermis, dermis and subcutaneous tissue in different incident wavelengths. Combining with the optical properties of skin tissue, the near infrared light propagation and distribution characteristics in skin tissue when subjected to different light source-detection distances from 1 000 nm to 1 900 nm were analyzed using Monte Carlo method. The results showed that photon path length and penetration depth increased with the increase of the source-detection distance, while normalized energy of diffuse light decreased with the increase of the source-detection distance. The source-detection distance was selected to be 0.45 mm. When the incident wavelength was 1 550 nm, the photon path length was 1.806 mm, the penetration depth was 0.467 mm, and normalized energy of diffuse light was 0.001 85. A kind of optical fiber detection structure was analyzed and designed according to Monte Carlo simulation results. The bifurcated fiber bundle was composed of 18 source fibers and 4 detection fibers, the distance of each fiber was 0.45 mm and just compactly adjacent. Finally, the diffuse light energy and illumination distribution collected by this optical fiber were simulated. Assuming that the incident light power was 1 W, the diffuse light power received by the detector was 0.598 mW. The results could provide a reference for the design of the portable detection spectra instrument.

For the measurement of a single drop in spray, the Debye series for scattering of a Gaussian beam by a spherical drop is used to study the light intensity distribution in the primary rainbow region and the influence of the Gaussian beam on the peak position of the intensity distribution. According to the intensity distribution calculated by the Debye series, the refractive index and size of drop are inversed, which prove that the Gaussian beam scattering could be used for drop measurements. And then, generalized Lorenz-Mie theory is used to compute the total light intensity which is used to inverse the refractive index and size of drop. And the influence of Gaussian beam position on the inversion of drop information is also investigated. For a Gaussian beam centered on the Descartes ray position, the absolute error of the refractive index is smaller than 2.38×10-4 and the absolute value of relative error of drop radius is smaller than 3.13% for drops with a radius between 200 and 1 000 μm. Compared with the rainbow technique using parallel beam, it possible to obtain a measurement area with high intensity and appropriate volume by using Gaussian beam. Therefore, the simultaneous occurrence of many drops in measurement area can be effectively avoided. And we could reduce the influence of multiple scattering between drops and increase the signal intensity.

Photometric data of satellites are related to many factors, such as satellite attitude, material and observation angle. The simulation calculation is often unable to obtain these factors, so that the simulation results cannot be compared with the measured results. Aiming at the problem that simulation results cannot be verified by measurement, the photometric data of the satellite in Satellite Tool Kit motion scene are obtained by experimental measurement and simulation calculation. In laboratory measurement, not only the space optical environment is simulated, but also the parallel analysis of satellite model attitude and satellite attitude in Satellite Tool Kit can be realized, which ensure the measurement accuracy. In simulation calculation, the bidirectional reflectance distribution function model based on the bidirectional reflectance distribution function data of the common space target material is used, and the influence of the wrinkled surface is taken into account, which improves the accuracy of the simulation. In addition, a three-axis stable satellite motion scene with phase angle variation range of 38°-98° is designed, the orbital parameters of the satellite are: semi major axis a=7 716.14 km, eccentricity e=0.001, orbit inclination i=58°, argument of perihelion ω=36°, longitude of ascending node Ω=345°, mean anomaly M=231.1°. When compare the photometric data obtained by experiment and simulation, the results show that two trends are same and the correlation coefficient is 0.69.

Based on the visual attention mechanism, a large number of subjective experiments were conducted to quantitatively study the chromaticity factors that affect the visual comfort of stereo images. First, the stereo salient map was obtained by combining disparity map and flat salient map, then the stereo salient map was optimized by using fuzzy membership and mask to get final salient image, and the eye tracker was used for the salient stereoscopic images to verify the rationality; then, a gradual approach method was used for subjective experiments to obtain experimental data, and after that remarkable comfortable color matching map and difference map of three-dimensional images in different scenes were obtained by a large number of subjective experiments. The experimental results show that the comfort and chromaticity interval of left and right view will vary with different scenes, and the maximum average chromaticity comfort difference of binocular vision can be 122.5°, that is to say, the chromaticity difference between left and right views should not be too large. The comfort and chromaticity range obtained can reflect the comfort of stereo images well, and provide the basis for judging the quality of stereoscopic information and more powerful technical support for stereo content making.